Metrology of features in images can be a difficult and lengthy process, especially on images obtained by a charged particle microscope. The difficulty and lengthy process may be due in part to noisy images that automatic imaging processing algorithms have difficulty processing, which leads to user manipulation and multiple steps. This may not be an issue when there are only a few images to analyze. However, in a manufacturing environment, such as the semiconductor industry, where large numbers of samples are imaged and require analysis, this process truly slows down needed analysis.
While some improvements to image processing have occurred over the years leading to improved metrology accuracy and efficiency, these improvements are not enough in today's semiconductor manufacturing climate, e.g., node size and throughput. As such, improvements to image process and metrology automation are desired across an industry.